Powder Metallurgy

The five-step process

Every powder-metallurgy (PM) part follows the same five-step recipe, with the last two repeated or substituted depending on the alloy and the duty cycle the part is meant to survive.

1. Atomize a stream of molten metal into 1–200 µm droplets, usually with a high-velocity gas or water jet, and let the droplets solidify in flight.
2. Blend the powder with any alloying additions (graphite for steels, Cu for infiltration, Co for hardmetals) and 0.5–1 % of a die-wall lubricant such as zinc stearate.
3. Press the blend in a rigid steel die at 200–1000 MPa to form a "green" compact — strong enough to handle, but only 80–90 % of theoretical density and roughly the strength of compacted chalk.
4. Sinter the green compact in a controlled inert or reducing atmosphere at 0.7–0.9 of the absolute melting temperature. Solid-state diffusion welds the particles together; lubricant burns off, oxides are reduced, and the part shrinks 0.5–3 % in each linear dimension.
5. Finish with optional sizing, machining, oil impregnation, copper infiltration, hot isostatic pressing (HIP), shot peening, or steam treatment — whatever the application needs.

This sequence produces a finished metal part that has never been molten in the cavity of a mold. Tolerance, microstructure, and even the porosity can be engineered at the powder stage rather than fought afterwards.

Step 1 — atomization

Powder is the raw material the rest of the process depends on, and the way it is made shapes what can be done with it.

Gas atomization directs a sonic-velocity jet of argon or nitrogen at a falling stream of molten metal. The gas strips off droplets that cool in flight, freezing into nearly spherical, smooth, low-oxide particles in the 10–150 µm range. Spherical particles flow well and pack tightly — essential for laser powder-bed fusion (LPBF) 3D printing — but the process is expensive: $50–200 per kg for Ti-6Al-4V or Inconel 718.

Water atomization uses high-pressure water jets (5–20 MPa) in place of gas. Cooling is so fast (10⁵–10⁶ K/s) that particles freeze irregular and angular before surface tension can round them. The water also oxidizes the surface — which is fine for most ferrous PM because the oxide layer is later reduced in the sintering furnace under hydrogen. Water-atomized iron and low-alloy steel powder sells for $5–15/kg and is the workhorse of structural PM.

Centrifugal atomization spins a molten droplet off the edge of a rapidly rotating disc or rod. The plasma rotating-electrode process (PREP), a centrifugal variant, melts the tip of a spinning titanium electrode with a plasma arc; droplets fly off into an argon-filled chamber and freeze as the cleanest, most spherical Ti powder available. Used for aerospace PM superalloy parts where every inclusion matters.

Chemical and electrolytic methods round out the family. Carbonyl iron (decomposing Fe(CO)₅ vapor) makes ultra-fine 1–10 µm spheres used in soft magnetic cores and pharmaceuticals. Electrolytic copper deposits a dendritic powder ideal for self-lubricating bronze bearings. Reduced tungsten powder is precipitated from ammonium-paratungstate solution and reduced under hydrogen.

Step 2 — blending

The press never sees pure metal powder. Three things go in together:

• Base powder (typically 95–98 % of the mass): water-atomized iron for structural PM steel; pre-alloyed gas-atomized stainless or superalloy for premium parts; tungsten carbide for cutting tools.
• Alloying additions: graphite (0.3–0.8 %) to make eutectoid steel, Cu or Ni for solid-solution strengthening, MoS₂ for self-lubrication, Co for WC-Co hardmetals.
• Pressing lubricant (0.5–1 %): zinc stearate, ethylene-bis-stearamide (EBS) waxes, or admixed proprietary blends. The lubricant reduces friction between the powder and the die walls during ejection; without it the green compact often cracks or seizes.

The mixture tumbles in a V-blender or double-cone mixer for ten to thirty minutes. Over-blending segregates fine and coarse particles; under-blending leaves alloy-content variations that bake into the final part. The empirical rule: blend until uniform on the scale of one die cavity, not longer.

Step 3 — die pressing and the green compact

Conventional PM uses a uniaxial double-action press: upper and lower punches squeeze powder against a rigid die wall in a single stroke. Multi-level parts (gears with hubs, bushings with flanges) use independently driven upper and lower punches plus core rods — modern presses run six independently controlled axes to hit final density uniformly across complex profiles.

Pressure is the single biggest lever. Soft iron and copper compact well at 200–400 MPa; pre-alloyed steel needs 600–800 MPa; tungsten-carbide cobalt hardmetal mixtures push 1000 MPa or more. Press tonnage scales with the projected area of the part — a 50 cm² gear blank at 700 MPa needs a 350-tonne press:

F_press = A_proj × p_compact
        = 50 cm² × 700 MPa
        = 50 × 10⁻⁴ m² × 7 × 10⁸ Pa
        = 3.5 × 10⁵ N  ≈  35 tonnes-force per cm² of projected area

Green density depends on pressure, particle shape, and lubricant content. Spherical gas-atomized powder packs to ~64 % of theoretical at zero pressure; angular water-atomized powder, 35 %. After 700 MPa pressing, both reach 85–90 % — enough green strength to demould and tray for sintering, but still fragile. A green compact has roughly the cohesion of a sugar cube and a tensile strength of about 5 MPa, three orders of magnitude weaker than the finished part.

Step 4 — sintering: where the physics happens

Sintering is the heart of powder metallurgy. The green compact enters a continuous belt or pusher furnace at 0.7–0.9 of the absolute melting temperature in a controlled atmosphere. Three things happen, in this order:

1. De-lubrication (200–600 °C). The pressing wax burns off in air or vaporises in nitrogen. If it doesn't leave cleanly the part traps soot and the alloy carbon goes off-target.
2. Reduction (600–1000 °C, hydrogen or dissociated ammonia). Surface oxides on the particles are reduced to metal, exposing clean surfaces that can bond.
3. Densification (0.7–0.9 T_m). Solid-state diffusion at the particle contacts forms necks, which grow into bonded grain boundaries. Pores round off and shrink. The mechanism is dominated by grain-boundary and lattice diffusion driven by the curvature differences between the necks (where surface tension pulls atoms in) and the pore walls.

The driving force is the reduction of total surface energy as the powder fuses. Quantitatively, the linear shrinkage during sintering follows Herring's scaling: time to reach a given density scales with the cube of particle radius for grain-boundary diffusion control. Cut particle size by 2× and sintering time drops 8×, which is why nano-powder routes (where they work) are so attractive.

Material	Sinter temperature	T / Tm	Atmosphere	Time
Sintered bronze (Cu-Sn)	820 °C	0.85	Dissociated ammonia	20–30 min
PM steel (Fe-C, Fe-Cu-C)	1120 °C	0.77	N₂ / H₂ (90/10)	20–40 min
Stainless 316L PM	1260 °C	0.80	H₂ or vacuum	30–60 min
WC-Co hardmetal	1380–1450 °C	liquid-phase	Vacuum	60–90 min
Tungsten	2200 °C	0.69	Hydrogen	several hours

The tungsten line is the clearest sales pitch for PM: tungsten melts at 3422 °C and is essentially impossible to cast. Pressing and sintering its powder at 2200 °C is the only economic route to filaments, X-ray targets, kinetic-energy penetrators, and the contacts inside every spark plug.

WC-Co hardmetal is the other PM standout. Tungsten carbide (m.p. 2870 °C) is sintered by adding 6–15 % cobalt, which melts at 1495 °C and forms a liquid binder phase that wets WC grains and pulls them together at 1380–1450 °C. The cobalt freezes around the carbide grains on cooling. The result — sintered tungsten carbide — sets the world standard for metal-cutting inserts and rock-drilling bits, with hardness up to 92 HRA (about three times that of hardened tool steel).

Step 5 — finishing and the role of HIP

As-sintered PM parts retain 5–10 % porosity. For most automotive PM components (gears, levers, sprockets) that is perfectly adequate: the part is sized in a coining die to hit final tolerance and then shipped. For higher-duty applications, the pores have to go.

Copper infiltration is the cheapest fix. A pellet of copper is placed on top of the green steel compact during sintering; molten Cu wicks into the pores by capillary action, raising bulk density to 99 % theoretical and adding about 10 wt% Cu. Used widely for con-rod and small-engine PM parts.

Hot isostatic pressing (HIP) is the gold-standard densification step. The sintered part is loaded into a pressure vessel, heated to roughly 0.7–0.9 T_m and pressurised with argon to 100–200 MPa for several hours. The isostatic gas pressure closes residual porosity by creep and diffusion bonding, lifting density to essentially 100 % theoretical. Fatigue life improves by an order of magnitude. HIP-grade PM superalloy turbine discs (Rene-95, IN-718) replaced cast-and-wrought discs in fighter and commercial engines starting in the 1970s and have remained the standard since.

Other common finishing steps:

• Oil impregnation for sintered bronze bushings: vacuum-soak the part in oil so capillarity fills the 20–30 % interconnected pores.
• Steam treatment for PM steel: hot steam grows a thin Fe₃O₄ layer on every internal surface, sealing the pores, raising compressive strength, and improving corrosion resistance.
• Surface densification (rolling or shot peening) closes pores in the top 0.5–1 mm of the surface — the zone where fatigue cracks initiate — without the full cost of HIP.
• Sinter-hardening: control the cooling rate at furnace exit so PM steel quenches directly from austenite to martensite, eliminating a separate heat-treat step.

Worked example: cycle time and cost for a PM transmission gear

A 35 mm diameter planetary pinion (projected area 10 cm², height 12 mm, mass 65 g) for an automatic transmission is a textbook PM part. Walk the numbers:

Press tonnage   F = A_proj × p     = 10 cm² × 700 MPa  ≈  70 tonnes
Press cycle     t_press            ≈  2 s  (modern double-action press, 30 spm)
Sinter belt     1120 °C × 30 min  in N₂/H₂  =  effective t_sinter  ≈  30 min
                (continuous belt: cycle time is set by part spacing, ~3 s)
Material cost   65 g × $1.50/kg water-atomized Fe + 0.5% graphite
                                  ≈  $0.10 per part
Tooling life    > 250 000 parts on a hardened-steel die set
Net effective cycle time          ≈  3 s per part on a hot belt
Per-part variable cost            ≈  $0.40–0.60 at high volume

Compare this with hobbing the same gear from wrought 8620 bar stock: roughly 90 seconds of machining per part, plus heat treatment, plus 30 % of the bar mass going to chips. The PM route wins on cost by roughly 3×, and the chips never exist in the first place. The trade-off — slightly lower fatigue strength due to residual porosity — is acceptable for planetary pinions running in oil with redundant load paths through three planet gears.

Advantages — what PM does that nothing else can

• Near-net shape. Sintered density 90–95 % of theoretical means almost no machining waste — typical PM part yields 95 % of the input powder as finished mass. Compare with 30–50 % yield for machined wrought stock.
• Refractory metals. Tungsten (m.p. 3422 °C), molybdenum (2623 °C), rhenium (3186 °C), tantalum (3017 °C) cannot be cast economically. PM is essentially the only industrial route.
• Hardmetals (WC-Co, TiC, mixed carbides). The cermet that dominates metal-cutting inserts and rock bits exists because liquid-phase sintering can wet and bond the carbide grains with a tough metallic binder.
• Engineered porosity. Oil-impregnated self-lubricating bronze bushings, gas-permeable bronze filters, and tunable osseointegration surfaces for orthopedic implants all rely on deliberately leaving the pores open.
• Compositions that cannot be cast. Cu-W contacts, Ag-Ni electrical contacts, Cu-graphite brushes — pairs of metals that are immiscible as liquids but co-exist happily as a sintered composite.
• Functionally graded materials. By layering different powders in the die or co-pressing soft and hard zones, you can put hard wear-resistant material on one face and tough ductile material on the other in a single press stroke.
• 3D-printing feedstock. Laser powder-bed fusion (LPBF, SLM) and electron-beam melting both consume gas-atomized PM powder. The entire metal-AM industry — including aerospace brackets, custom implants, and Relativity Space's rocket engines — runs on PM-grade powder.

Where you actually find sintered metal parts

Application	Material	Why PM wins
Tungsten-carbide cutting inserts	WC-6 to WC-15 Co	Only practical route to a hard, tough cermet
Automatic-transmission planetary pinions	Fe-Cu-C PM steel	Net shape + sinter-hardening; no machining
Self-lubricating bushings (fans, small motors)	Bronze 90Cu-10Sn, oil-impregnated	Engineered porosity stores lubricant
Cordless-tool gearbox sprockets	PM steel + steam-treated	Quiet, light, no oil leakage from sealed pores
Magnetic cores for stepper motors	Pure iron PM, soft-magnetic composite (SMC)	3D flux paths in a single pressing
Aerospace turbine discs (IN-100, Rene-95)	Gas-atomized superalloy + HIP	Cleaner microstructure than cast-and-wrought
Orthopedic implants	Ti-6Al-4V PM + porous coating	Bone in-growth into engineered porosity
Tungsten X-ray targets, kinetic-energy penetrators	Pure or W-Ni-Fe heavy alloy	Cannot cast tungsten economically
Spark-plug centre electrodes	Cu-W or pure W tips	Erosion resistance + thermal conductivity
Metal-3D-printing feedstock	Gas-atomized Ti, Al, Inconel, stainless	Sphericity, flowability, low oxygen

Limits, failure modes, and pitfalls

• Size cap. Press tonnage scales with projected area, so conventional PM parts top out around 5 kg and a few hundred cm² of press face. Bigger parts go through cold isostatic pressing (CIP), hot pressing, or HIP-of-canned-powder routes — slower and more expensive.
• Residual porosity hurts fatigue. Even at 95 % density, the remaining pores act as stress concentrators. Endurance limits in as-sintered PM steel run 50–70 % of equivalent wrought steel. HIP, infiltration, or surface densification recover most of the gap, at a cost.
• Powder handling is hazardous. Fine reactive metal powders (Al, Mg, Ti, Zr) are pyrophoric and dust-explosion risks. Atomization plants and 3D-printing powder handling require nitrogen blanketing, conductive flooring, and stringent housekeeping. Industrial PM titanium fires are not rare.
• Lubricant burnout creates emissions. The first zone of a sintering furnace incinerates ~1 % of the part mass as zinc stearate fumes. Inadequate ventilation deposits the residue on furnace internals and contaminates downstream zones.
• Alloy segregation during blending. Density and size differences cause graphite to settle and Cu to float in poorly engineered blenders, drifting the average composition of every fifth part. The fix is admixed pre-coated binder-treated powder, which fixes the graphite to the iron particle surface.
• Shrinkage variability. Sintering shrinks the part 0.5–3 % in each linear dimension. Variability in that shrinkage is the largest source of dimensional error; modern PM tolerances are quoted "after sintering, before sizing" for that reason.
• Anisotropy from die pressing. Density gradients along the pressing axis are unavoidable in tall parts — the powder column friction reduces transmitted pressure with depth. Multi-action presses with floating dies and split punches push the limit, but very tall parts (height > 3× diameter) are simply not made by conventional PM.

Variants and adjacent processes

• Metal injection moulding (MIM). PM's offshoot for very small, complex parts. Powder (fine, < 20 µm) is mixed with 35–45 vol% polymer binder, injection-moulded like plastic, debound by solvent and thermal cycles, and sintered. Hearing-aid components, watch cases, gun-trigger parts.
• Cold isostatic pressing (CIP). Powder is loaded into an elastomer bag and pressurised hydrostatically in water to 200–600 MPa. Uniform green density in all directions, no die-wall friction. Used as a pre-press for large or complex shapes, including titanium aerospace blanks.
• Hot isostatic pressing (HIP) of canned powder. Powder in a steel can is HIPed at temperature and pressure simultaneously — full densification from powder in one step, no separate sinter. The route for large near-net superalloy preforms and titanium aerospace structures.
• Laser powder-bed fusion (LPBF / SLM) and electron-beam melting (EBM). Metal additive manufacturing — selective melting of PM powder layer by layer. Same atomized feedstock, no die, no fixed tooling; pays off only for geometries impossible by conventional PM.
• Direct energy deposition (DED, LENS). PM powder blown through a nozzle into a focused laser or arc; used for repair (e.g. turbine-blade tips) and gradient deposits.
• Spark plasma sintering (SPS). Powder is consolidated under uniaxial pressure while a pulsed direct current passes through it — heating times of minutes rather than hours, with very fine microstructures. Mostly a lab process; commercial throughput is limited.

Brief history

The oldest surviving PM artifact is the Delhi Iron Pillar (5th century AD): pre-reduced sponge iron blooms hammer-welded into a corrosion-resistant 7-tonne column. The modern industry began with William Coolidge's tungsten lamp filaments at General Electric in 1909 — the first commercial product no melt-based process could deliver. Karl Schröter at Krupp patented the WC-Co hardmetal in 1923, launching the cemented-carbide cutting-tool industry that still dominates metal cutting a century later. Oilite self-lubricating bronze bearings followed in 1930. PM iron parts entered automotive production in the 1940s; by the 1990s an average North American automobile contained 15 kg of sintered components. Gas-atomized superalloy turbine discs are the late-20th-century triumph; metal 3D printing on PM powder is the early-21st.

Common pitfalls

• Treating PM steel as wrought steel. Tensile strength is similar; fatigue endurance limit and notch sensitivity are not. Design rules and S-N curves come from PM-specific datasheets (MPIF Standard 35), not from wrought-steel handbooks.
• Ignoring sintering shrinkage in tool design. Dies must be cut oversize by the average shrinkage so the post-sinter part hits drawing tolerance. Get the shrinkage allowance wrong and every part is rejected.
• Confusing green density with sintered density. Green density (after pressing) is roughly 85 %; sintered density (after furnace) is 90–95 %. Quoting one when the customer expects the other is the most common source of arguments in a PM supply contract.
• Skipping reduction atmosphere on stainless or ferritic alloys. Oxygen pickup during sintering forms chromium-rich oxides that pin grain boundaries and embrittle the part. Vacuum or hydrogen sintering is non-negotiable for stainless PM.
• Forgetting that infiltrated copper changes the alloy. Cu-infiltrated PM steel contains ~10 wt% Cu — close to a bronze in some properties. Heat-treat response and weldability are not what you'd expect from a Cu-free steel of the same nominal composition.
• Pressing too tall a part on a single-action press. Die-wall friction makes the top of a tall column denser than the bottom. The fix is double- or multi-action pressing, not more tonnage.